Analytica Chimica Acta 989 (2017) 112e120

Contents lists available at ScienceDirect

Analytica Chimica Acta

journal homepage: www.elsevier.com/locate/aca

Ultra-high performance supercritical fluid chromatography-mass

spectrometry procedure for analysis of monosaccharides from plant
gum binders
s Pluha
Volodymyr Pauk, Toma  cek, Karel Lemr*
cek, Vladimír Havlí
Regional Centre of Advanced Technologies and Materials, Department of Analytical Chemistry, Faculty of Science, Palacký University in Olomouc, 17.
listopadu 12, 771 46 Olomouc, Czech Republic

h i g h l i g h t s g r a p h i c a l a b s t r a c t

A new SFC-based method for analysis

of monosaccharides was developed.

SFC analysis of plant gum binders is
reported for the first time.

Fast sample preparation and separa-
tion contribute to significant time
and labor saving.

Plant gum binder was detected in
watercolors applied onto interfering
paper support.

a r t i c l e i n f o a b s t r a c t

Article history: The ultra-high performance supercritical fluid chromatography-mass spectrometry (UHPSFC/MS) pro-
Received 9 March 2017 cedure for analysis of native monosaccharides was developed. Chromatographic conditions were
Received in revised form investigated to separate a mixture of four hexoses, three pentoses, two deoxyhexoses and two uronic
19 June 2017
acids. Increasing water content in methanol modifier to 5% and formic acid to 4% improved peak shapes
Accepted 16 July 2017
Available online 21 July 2017
of neutral monosaccharides and allowed complete elution of highly polar uronic acids in a single run. An
Acquity HSS C18SB column outperformed other three tested stationary phases (BEH (silica), BEH 2-
ethylpyridine, CSH Fluoro-Phenyl) in terms of separation of isomers and analysis time (4.5 min).
Keywords:
Supercritical fluid chromatography
Limits of detection were in the range 0.01e0.12 ng mL1. Owing to separation of anomers, identification of
Mass spectrometry critical pairs (arabinose-xylose and glucose-galactose) was possible. Feasibility of the new method was
Microwave-assisted hydrolysis demonstrated on plant-derived polysaccharide binders. Samples of watercolor paints, painted paper and
Monosaccharide three plant gums widely encountered in painting media (Arabic, cherry and tragacanth) were decom-
Plant gum posed prior the analysis by microwave-assisted hydrolysis at 40 bar initial pressure using 2 mol L1
Watercolor trifluoroacetic acid. Among tested temperatures, 120  C ensured appropriate hydrolysis efficiency for
different types of gum and avoided excessive degradation of labile monosaccharides. Procedure recovery
tested on gum Arabic was 101% with an RSD below 8%. Aqueous hydrolysates containing mono-
saccharides in different ratios specific to each type of plant gum were diluted or analyzed directly.
Filtration of samples before hydrolysis reduced interferences from a paper support and identification of
gum Arabic in watercolor-painted paper samples was demonstrated. Successful identification of pure
gum Arabic was confirmed for sample quantities as little as 1 mg. Two classification approaches were
compared and principal component analysis was superior to analysis based on peak area ratios of
monosaccharides. The proposed procedure using UHPSFC/MS represents an interesting alternative which

* Corresponding author.
E-mail address: Karel.Lemr@upol.cz (K. Lemr).

http://dx.doi.org/10.1016/j.aca.2017.07.036
0003-2670/© 2017 Elsevier B.V. All rights reserved.
 V. Pauk et al. / Analytica Chimica Acta 989 (2017) 112e120 113

can compete with other chromatographic methods in the field of saccharide analysis in terms of speed,
sensitivity and simplicity of workflow.
© 2017 Elsevier B.V. All rights reserved.

existing methods used for binder analysis are labor- and time-
Abbreviations consuming multistep procedures and have numerous disadvan-
tages. High performance anion-exchange chromatography (HPAEC)
HPAEC high performance anion exchange chromatography with pulsed amperometric detection required clean-up of hydro-
MAH microwave-assisted hydrolysis lysates on a cation-exchange column, eluate evaporation, residue
PCA principal component analysis redissolving, 50 min runtime and additional electrode surface
UHPSFC ultra-high performance supercritical fluid cleaning/restoring steps [8]. Man and Xyl peaks were partially
chromatography overlapped. CE methods with indirect UV detection, despite being
Ara arabinose fast, necessitated overnight lyophilization of hydrolysates and
Fru fructose highly alkaline background electrolyte (pH 12) [4,12]. The methods
Fuc fucose did not separate pairs Man-Xyl and Ara-Glc. High concentration of
Gal galactose sodium hydroxide in the HPAEC and CE methods hinders applica-
GalA galacturonic acid tion of MS to distinguish these critical pairs. GC-MS separations are
Glc glucose usually long (almost 1 h) and require extensive sample treatment,
GlcA glucuronic acid such as hydrolysate clean-up on cation-exchange resin, drying,
Man mannose conversion to diethyl dithioacetals and lactones with following
Rha rhamnose trimethylsilylation and final solvent exchange [10,11,13]. Alterna-
Rib ribose tively, monosaccharides can be converted to methoximes and
Xyl xylose acetylated [10]. A simplified GC-MS procedure based on thermally
assisted alkaline hydrolysis and methylation with pyrolysis did not
provide a complete monosaccharide profile [7,9]. Uronic acids were
not detected and neutral monosaccharides epimeric on C-2 were
not distinguished.
1. Introduction The described methods mostly report analysis of pure materials
or paint applied to glass, clay or plaster (mural paintings), i.e. gum/
Identification of plant gum-based binders is one of the most paint layers were relatively easily detached from the matrix. Wood
interesting and demanding tasks in the field of saccharide analysis. and paper support were also considered but only in combination
Plant gums consist of complex branched polysaccharides. They with gouache or tempera [6]. The most expected and the most
have been extensively used in water-based painting media challenging gum application scenario, watercolors on paper, was
(watercolor, gouache, gum tempera, metallo-gallic inks) as well as assessed with only one method [4]. Painted paper was first treated
in dry painting materials (pastels, pencils, charcoal) [1,2]. Gum in boiling water, filtered and then hydrolyzed. A watercolor paint
Arabic (Acacia sp.), cherry (Prunus avium) and tragacanth (Astra- itself cannot be detached from paper easily and must be analyzed
galus sp.) were commonly utilized in artworks and were applied as only as a scrape or cutout containing huge excess of paper-derived
binding media for pigments in Egyptian ointments used for monosaccharides since it is applied in very thin layers and pene-
mummification, mural paintings in Christian catacombs, paintings trates deeper into the paper structure. On the contrary, tempera
on silk and manuscript illuminations in the Middle Ages [1]. and other denser media usually form a film on the surface of a
Additionally, plant gums have been used as components of var- substrate and are only partially absorbed inwards. It is known that
nishes and glues. Identification of a particular binder is useful for tempera layers can be easily scraped down to the bare gesso/
restoration purposes, evaluation of object authenticity and its primer, which was often used to cover rugged surfaces before
possible provenance. applying paint.
Plant gums are mainly identified detecting specific mono- Hence, a technique capable of fast separation of mono-
saccharides, such as arabinose (Ara), xylose (Xyl), fucose (Fuc), saccharides without derivatization or excessive sample pre-
rhamnose (Rha), glucose (Glc), galactose (Gal), mannose (Man) and treatment would be highly valuable. Compatibility with MS
two sugar acids, glucuronic acid (GlcA) and galacturonic acid (GalA) detection would favor identification of different monosaccharide
[1,3e6]. Relative ratio of monosaccharides [7], principal component classes and method selectivity in presence of interfering matrices.
analysis (PCA) [8e10] and cluster analysis [9] improve discrimi- Modern ultra-high performance supercritical fluid chromatography
nation between plant gums. A prolonged treatment (up to 24 h) (UHPSFC) may be a rapid and effortless alternative for raw mono-
with heated strong acid such as HCl or H2SO4 is needed for com- saccharide identification and binder analysis.
plete hydrolysis of glycosidic bonds but degradation of labile SFC represents a substitute to NP-LC offering fast analysis and
monosaccharides may occur [1]. Microwave-assisted hydrolysis environmental benefits [14]. Applicability of SFC to polar com-
(MAH) carried out at higher temperature (100e120  C) speeds up pounds was traditionally judged according to their solubility in
the treatment, provides higher monosaccharide yield and avoids methanol or less polar solvent [15] or their Log P (values > 2 were
contamination [8,10,11]. Nevertheless, the cited MAH method was considered to be reasonable) [16]. Therefore, this technique was not
optimized using only gum Arabic containing four monosaccharides seriously considered as a method of choice for native mono-
and hydrolysis efficiency for other gums/monosaccharides saccharides (Log P < 2 [17]). However, the range of SFC analytes
remained undetermined. was significantly extended using mobile phase modifiers and ad-
Monosaccharides can be separated by various techniques but ditives making it complementary to RP-LC or even alternative to
114 V. Pauk et al. / Analytica Chimica Acta 989 (2017) 112e120
HILIC [18e21]. A few papers on packed column SFC-ELSD of car-
bohydrates including native monosaccharides and sugar alcohols
were published in mid-90s and reviewed in Ref. [22]. A number of
substances was investigated using CO2/methanol mobile phase.
Resolution of pairs Ara-Xyl and Gal-Glc, deduced from the retention
times on four columns (Zorbax CN, m Bondapak CN, LiChrospher
diol and RSil NO2), was not achieved. The fifth column, LiChrosorb
CN did not resolve Glc and Man. Polar Zorbax Sil and non-polar
ZorbaxTMS phases were tested with 0, 4 and 8% of water in
methanol as a modifier [23]. Water caused increase in retention of
saccharides and anomer separation for Xyl and Gal. Authors sug-
gested partition mechanism due to increased solvation of a sta-
tionary phase by the hydroorganic modifier. Based on the retention
factors, separation of monosaccharides improved comparing to
pure methanol but pairs Ara-Xyl and Gal-Glc remained unresolved
on both columns. Most of these works dealt only with pure stan-
dards. The only method with a real-world application, analysis of
Glc, fructose (Fru) and succrose in tobacco leaves, utilized nonpolar
Zorbax ODS and Zorbax TMS bonded silica phases [22]. Other
saccharides crucial for plant gum analysis, such as Fuc, GlcA and
GalA, were not included in the mentioned works on SFC.
The primary goal of our study was to develop a new SFC method
for analysis of monosaccharides and prove its suitability for
demanding real-world applications using plant gums as a model
example. In our experiments, we included all monosaccharides
encountered in plant gums used in painting media and two sugars
that are not derived from plant gums but are widespread in nature
and can be present as contaminants (ribose (Rib) and Fru). Influ-
ence of high-pressure MAH temperature was investigated to
effectively hydrolyze glycosidic bonds and to avoid degradation of
labile components at the same time. In addition to analysis of raw
materials and paint with non-interfering matrix, the challenging
scenario, watercolors applied onto paper, was evaluated. We
compared two methods for classification of plant gums based on
relative ratios of monosaccharides or PCA. To our best knowledge,
this is the first report on utilization of UHPSFC-MS for analysis of
plant-derived gum binders.
2. Experimental
2.1. Chemicals
Carbon dioxide 4.8 grade (99.998%) was provided by SIAD
(Prague, Czech Republic). Methanol and acetonitrile, both LC-MS
grade, ethanol HPLC grade, formic acid (95%), acetic acid
(99.7%), trifluoroacetic acid (TFA) (99.0%), sodium acetate
(99.0%) ammonium formate (99.0%) and ammonium hydroxide
(25% ammonia in water) were purchased from Sigma-Aldrich
(Prague, Czech Republic). Water LC-MS grade (Merck, Darmstadt,
Germany) was used as mobile phase modifier in SFC. Water for
sample preparation was purified by Milli-Q system (Millipore,
Mollsheim, France). D-ribose, D-xylose, L-arabinose, L-rhamnose
monohydrate, D-fructose, D-mannose, D-glucose, D-galactose, D-
glucuronic acid and D-galacturonic acid monohydrate (all 97.0%)
were purchased from Sigma-Aldrich (Prague, Czech Republic). L-
fucose (97.0%) was supplied by Acros organics (Geel, Belgium).
Structures of the investigated saccharides are shown in Fig. 1. Plant
gums, Arabic (kibbled), cherry (kibbled), tragacanth (powder) were
obtained from Kremer Pigmente (Aichstetten, Germany). Winsor &
Newton Cotman watercolor half pans of Alizarin crimson, Hooker's
Green and Prussian Blue and paper sheets were bought at a local art
supplies shop in Olomouc, Czech Republic.
Fig. 1. Structures of the investigated saccharides.
2.2. Sample preparation
Monosaccharide stock solutions were prepared in water at
10 mmol L1 concentration and stored in a refrigerator at 18  C.
Working solutions contained 20 mmol L1 of each monosaccharide
standard and were prepared fresh. Lumps of gum Arabic were
grounded in a ceramic mortar. Lumps of cherry gum were chopped
with a stainless steel scalpel. Gum tragacanth was used without
homogenization. Watercolors were scraped off half pans with a
scalpel. Samples of paper and paper painted with a thin layer of
watercolor were cut out with steel scissors. Samples of gums, wa-
tercolors, paper and paper painted with Prussian Blue were
weighed into glass HPLC vials (10 mg, approx. 5  5 mm), 1.0 mL of
2 mol L1 TFA in water was added and sonicated for 10 min in an
Elma S60H ultrasonic bath (Elmasonic, Singem, Germany) in sweep
mode. Samples were hydrolyzed in open vials using a temperature
controlled microwave digestion unit UltraWAVE (Milestone, Sir-
isole, Italy) at 40 bar initial pressure of nitrogen. The two-step
microwave program consisting of a 3 min temperature gradient
from room temperature to 100, 120 or 140  C maintained for 20 min
(500 W), with subsequent cooling to 40  C was applied (partially
adopted from Ref. [8]). Three samples of each gum and watercolors
and one sample of paper and painted paper were digested at each
temperature setting. Hydrolysates were filtered through 13 mm,
0.22 mm nylon LUT syringe filters (Cronus, Gloucester, UK). Gum
hydrolysates were diluted with deionized water 500 times prior
analysis. Watercolor, paper and painted paper hydrolysates were
diluted 100 times.
Additional experiments included a filtration step after sonicat-
ion in order to remove interfering paper support before the MAH
procedure and keep only a soluble gum binder. Samples were hy-
drolyzed at 120  C and analyzed undiluted to increase sensitivity.
Procedure sensitivity and recovery was evaluated using 1.0 mL of 1,
5, 10 and 20 mg mL1 solutions of gum Arabic in 2 mol L1 TFA.
Recovery was calculated as a sum of all monosaccharides found in
hydrolysates divided by the sample weight.
2.3. Instruments and MS detection
An Acquity UPC2 system with an Isocratic Solvent Manager
module coupled to a Xevo TQS triple quadrupole mass spectrom-
eter with a Z-spray electrospray source (all Waters, Manchester,
UK) was used. Control of the instruments and data acquisition was
performed using Waters MassLynx 4.1 software (Waters, Man-
chester, UK).
 V. Pauk et al. / Analytica Chimica Acta 989 (2017) 112e120
Table 1
SRM settings for analyzed monosaccharides.
Substance Formula Mass, Da Precursor ion
Pentoses C5H10O5 150.05283 323.00
Deoxyhexoses C6H12O5 164.06847 350.91
Hexoses C6H12O6 180.06339 382.97
Uronic acidsa C6H10O7 194.04265 192.88
Fuc C6H12O5 164.06847 181.97
Gal, Man C6H12O6 180.06339 163.00
Gal, Man, Fru C6H12O6 180.06339 198.00
a
Negative ion mode.
Source conditions were tuned during direct infusion of
20 mmol L1 standard solutions (5 mL min1) with make-up liquid
and set for positive mode as follows: capillary voltage 2.75 kV;
source offset 50 V; desolvation gas temperature 300  C; source
temperature 150  C; cone gas flow 150 L h1; desolvation gas flow
800 L h1; nebulizer gas pressure 3 bar. For negative mode capillary
voltage was 2.00 kV, source offset 30 V, other settings as above.
Continuous polarity switching was used during the method
development and sample analysis.
Make-up liquid consisted of 1 mmol L1 sodium acetate in water/
methanol (1:1, v/v) mixture and was delivered at 0.4 mL min1.
Selected reaction monitoring (SRM) transitions (Table 1) were
chosen and tuned in manual mode by means of Waters IntelliStart
software (Waters, Manchester, UK). Dwell time was 16 ms, inter-
channel delay was 3 ms, polarity switch delay was 20 ms.
2.4. Chromatographic conditions
Four Acquity UPC2 stationary phases were tested: BEH (silica)
1.7 mm, BEH 2-EP (2-ethylpyridine) 1.7 mm, CSH Fluoro-Phenyl
1.7 mm and HSS C18SB 1.8 mm (all from Waters, Manchester, UK). All
columns possessed the same dimensions 100  3 mm i.d. and were
used with in-line pre-column filters.
Effect of several modifiers and additives including methanol,
water, ethanol, acetonitrile, formic acid, acetic acid, TFA, ammo-
nium hydroxide and ammonium formate was investigated. For
each stationary phase two column temperatures (35 and 60  C)
were tested. Furthermore, 30  C was tested on BEH and C18SB.
Automatic backpressure regulator (ABPR) was set to 2000 psi
(138 bar). Mobile phase flow rate was 2 mL min1. Elution program
for column testing was set as follows: initial e 0% modifier;
8 mine25% modifier; 9e10 mine0% modifier. Pressure limit was set
at 6000 psi (414 bar). Strong and weak needle wash consisted of
water/methanol 9:1 (v/v) and 1:1(v/v), respectively.
Chromatographic conditions were evaluated separately for each
column. Final analysis was carried out on the C18SB column. ABPR
pressure was 2000 psi (138 bar), column temperature was 35  C.
Modifier consisted of methanol/water/formic acid (91:5:4, v/v/v).
Mobile phase flow rate was 2.0 mL min1. Elution program was set
as follows: initial e 0% modifier; 5 mine30% modifier;
5.5e7 mine0% modifier.
Mixture containing all monosaccharides and uronic acids
(20 mmol L1 each) or samples were injected (2 mL). Blanks were run
after each change of modifier or temperature for better column
equilibration and after each set of samples to control possible
system contamination. Data processing was performed using Tar-
getLynx software (Waters, Manchester, UK). When necessary,
integration range was corrected manually using base-to-base
method. PCA was performed by means of OriginPro 2015 soft-
ware (OriginLab Corporation, Northampton, MA, USA) using
115
Cone voltage, V Product Collision energy, V
ion
15 173.00 10
20 186.99 10
22 202.99 10
26 112.87 12
88.90 10
59.00 16
16 74.96 12
26 91.00 10
20 127.00 16
relative chromatographic peak areas of monosaccharides as
variables.
3. Results and discussion
3.1. Method development
The observed monosaccharide precursor ions were in agree-
ment with the literature [24]. Neutral sugars produced sodium
adducts [MþNa]þ and sodiated dimers [2MþNa]þ. [MþNa]þ ions
gave no fragmentation products except m/z 23 (Naþ) of low in-
tensity. [2MþNa]þ ions lost one monosaccharide moiety and
[MþNa]þ products were detected in MS/MS experiments. There-
fore, the sodiated dimers were chosen for SRM transitions (Table 1).
Additionally, Gal and Man in-source dehydration products m/z 163
[M-H2OþH]þ gave transitions to m/z 91. Fuc, Gal, Man and Fru
produced ammonium adducts [MþNH4]þ valuable for comple-
mentary confirmation. Signals of hexoses in negative mode [M-H]-
and [2M-H]- were roughly 20 times weaker than [2MþNa]þ. Uronic
acids formed [MþNa]þ adducts in positive mode but their frag-
mentation spectra were poor (only m/z 23). Thus, their deproto-
nated ions were used as precursors for SRM in negative mode.
Finally, continuous polarity switching was used for detection of all
substances in one chromatographic run. Stable sodium adducts
were observed without deliberate introduction of salts into the
make-up liquid or mobile phase modifier. Nevertheless, sodium
acetate (1 mmol L1) was added into the make-up liquid to prevent
possible signal instability and reproducibility issues.
Supercritical fluid chromatography was used as a technical term
throughout the text, though rather subcritical conditions existed in
chromatographic system for used mobile phase compositions.
Addition of the appropriate modifier was crucial for successful
separation of polar analytes. Changes of mobile phase composition
influenced chromatographic behavior of analytes similarly on all
stationary phases. Pure methanol did not elute uronic acids in
reasonable time, separation of individual hexoses, pentoses and
deoxyhexoses was unacceptable and their peaks were distorted.
Addition of 2% water and increasing its content to 5% significantly
improved chromatograms. Its higher content caused stronger
retention of neutral monosaccharides on all phases except Fluoro-
Phenyl. Further increase of water content in modifier (7e10%)
induced pressure instability and system overpressure at higher
percentage of modifier, most likely due to phase separation. It is not
surprising, since supercritical temperature and pressure of water
(647 K, 220 bar) are much higher than those of CO2 (304 K, 74 bar)
[25].
Mixture of methanol and water did not have enough strength to
elute uronic acids. Additives such as 20 mmol L1 ammonium hy-
droxide and ammonium formate, alone and in combination with
water, worsened peak shapes while formic acid produced sharper
116 V. Pauk et al. / Analytica Chimica Acta 989 (2017) 112e120
peaks. It is known that alkaline pH and presence of amines accel-
erates interconversion of anomers [26]. In our case used concen-
tration of ammonium additives caused merging of anomers but was
not sufficient to produce narrow peaks. Mixture of 5% water and 2%
formic acid in methanol lead to anomer splitting and provided the
best peak shapes and separation of individual monosaccharides.
However, uronic acids were not completely eluted during the
chromatographic run. Increasing content of formic acid in modifier
to 4% maintained separation of monosaccharides and allowed to
elute uronic acids completely except from the 2-EP column, where
strong electrostatic interaction between protonated stationary
phase and acid anions could occur. It is worth noting that at the
highest point of modifier gradient (30%) content of formic acid in
the mobile phase was only 1.2%. Comparing to neutral mono-
saccharides, uronic acids peaks were much broader and asym-
metric, apparently due to influence of dissociation. For its effective
suppression, more acidic environment would be required since pKa
values in water, 3.20 and 3.48 for GlcA and GalA, respectively [27],
are comparable with formic acid value 3.75 [28]. Weaker acetic acid
(2e4% in modifier) was not efficient to elute uronic acids but
rendered similar separation of other monosaccharides. 0.1% TFA
with 5% water in methanol showed results comparable with 2%
formic acid concerning retention, peak shapes and separation but
MS signal of late-eluting compounds was significantly decreased,
roughly 3 times for hexoses and 10 times for uronic acids (data not
shown). Replacement of methanol with ethanol as a modifier in
combination with 5% water and 2% formic acid rendered compa-
rable separation but several drawbacks were observed (see Sup-
plementary electronic material for more details, Fig. S1). For
example Gal produced distorted peaks and peak height dropped
one order of magnitude. Acetonitrile with 5% water and 2% formic
acid as a modifier failed to elute any of the studied substances.
Column temperature significantly affected interconversion of
anomers. At 35  C most sugars (Rha, Fuc, Ara, Xyl, Glc and Gal)
provided two peaks corresponding to a and b anomers, depending
on a stationary phase. Separation of sugar anomers might be seen
as a drawback [23]. In our case, for some monosaccharides one of
the anomers overlapped with another sugar (Ara-Xyl, Glc-Gal) but
the second could be used for identification (Fig. 2a). At 60  C
anomers merged into one peak and identification of certain sugars
became impossible (Fig. 2b). Lower temperature (30  C) did not
Fig. 2. Influence of temperature on separation of a mixture of standards on C18SB,
gradient: 0e25% methanol (5% H2O, 2% HCOOH) in 9 min, 2.0 mL min1: a) 35  C; b)
60  C.
result in further improvement of separation on the most suitable
stationary phases (BEH and C18SB).
Retention of monosaccharides on all stationary phases was
related to the number of hydroxy groups. Pentoses and deoxy-
hexoses were eluted first, followed by hexoses and, finally, uronic
acids (for more details see Fig. S2). This elution order is in agree-
ment with the literature [22] and strongly resembles a typical
behavior of sugars in HILIC mode [26]. Uronic acids were not
separated and were eluted as broad bands due to their high polarity
and presence of deprotonated forms. The presented SFC method
may not be a method of choice for quantitative analysis of these
acids but at least their presence might be confirmed.
Since neither stationary phase completely separated all mono-
saccharides, backpressure, mobile phase flow rate and gradient
were tuned for each column. Three critical pairs common to all
stationary phases were observed: Fuc-Rha, Ara-Xyl and Gal-Glc.
Under the most suitable conditions C18SB and BEH clearly out-
performed other stationary phases and allowed to identify each
monosaccharide (Fig. 3). C18SB provided better separation of Fuc
and Rha while BEH baseline resolved Fru and Man. 2-EP separated
neither deoxyhexoses nor Xyl and Ara. As it was mentioned, uronic
acids were not eluted from this column. Fluoro-Phenyl did not
render separation of Xyl and Ara, deoxyhexoses were separated
only partially. The C18SB column was chosen for analysis of real
samples due to better separation of monosaccharides relevant to
plant gums and faster runtime.
Significant attention has been paid to the injection solvent in
SFC since it can cause a detrimental peak distortion, especially for
early-eluting compounds, and thus influence the overall perfor-
mance of the chosen method [29]. Generally, low elution strength
solvents such as alkanes or at least their blends with higher polarity
solvents are recommended for SFC. Methanol or chloroform-
methanol were urged for analysis of carbohydrates [22]. Since
monosaccharides and uronic acids are insoluble in non-polar sol-
vents and only sparingly soluble in alcohols, water/methanol
mixture (1:1, v/v), pure water and water with 2 mol L1 TFA were
tested. All solvents provided equally good results on C18SB and BEH
columns, high content of TFA was not detrimental to separation,
which allows for direct analysis of filtered hydrolysate. This fact
might be explained by a small injection volume (2 mL), sufficient
retention of analytes and decreased interaction of polar sample
solvent with a stationary phase when large amount of water-rich
modifier is used comparing to an injection into pure CO2.
3.2. Microwave-assisted hydrolysis
On the contrary to the published methods, we utilized a high
pressure MAH system where samples can be processed at high
temperatures (>100  C) under inert atmosphere without boiling,
ensuring better reproducibility and avoiding sample loss or
contamination. The original MAH method was developed using
gum Arabic, which contains only four monosaccharides [8]. Here
temperature effect on individual monosaccharides content in three
plant gums and watercolors was tested. Higher amount of indi-
vidual monosaccharides can be released from a polysaccharide
sample applying higher temperature but certain monosaccharides
(Ara, Xyl, Rha, Fuc) easily undergo degradation. A compromise
temperature setting should be searched for. The highest peak areas
for labile sugars were obtained at 100  C for all tested materials
(Fig. 4). Hexoses, especially Man, were released better at higher
temperatures. No Man trace was evident in the cherry gum hy-
drolysate obtained at 100  C. Uronic acids showed the highest
response at 120  C. It is worth noting that at 140  C their content in
gum tragacanth hydrolysate decreased dramatically but remained
almost constant in gum Arabic and cherry gum hydrolysates.
 V. Pauk et al. / Analytica Chimica Acta 989 (2017) 112e120
Fig. 3. Optimized separation of standards mixture, 35  C: a) C18SB, gradient: 0e30% modifier in 5 min, 2.0 mL min1; b) BEH, gradient: 5e20% modifier in 9 min, 2.5 mL min1.
Modifier: methanol/water/formic acid (91:5:4, v/v/v).
According to the literature, gum tragacanth contains GalA while
cherry and Arabic gums contain GlcA [2,10]. For MAH lasting
20 min, 120  C represented an optimal setting. Degradation of
monosaccharides was tested with the standard solution
(20 mmol L1). For most monosaccharides apparent changes were
lower than the precision of the chromatographic method itself. The
most significant decomposition occurred for Fru (90%) and Rib
(32%) which are not encountered in gum binders, followed by
uronic acids (22%) and Xyl (20%). MAH procedure recovery tested
on gum Arabic at four levels (1, 5, 10 and 20 mg mL1) was 101% with
an RSD below 8%.
3.3. Analysis of plant gums, watercolors and paper
Gum Arabic contained Gal, Ara, Rha and small amount of uronic
acids (Fig. S3). Cherry gum contained Ara, Gal and small amount of
Xyl, Rha, Man and uronic acids. Gum tragacanth besides major
constituents Ara and Gal contained significant amount of Glc, Xyl,
Fuc, small amount of Rha and uronic acids. Neither Fru nor Rib were
found in gum hydrolysates.
To test the real-world performance of our method, commercial
high-grade watercolors were analyzed intact and as applied onto
paper. Gal, Ara and Rha, specific to gum Arabic, were found in all
watercolor hydrolysates. Hydrolysates of Hooker's green addition-
ally contained high amount of Glc. According to the manufacturer,
not only gum Arabic but also dextrin is used as a watercolor binder,
which explains Glc presence. Since only one of three paints con-
tained Glc, this saccharide was omitted from watercolors compar-
ison. Unfiltered paper hydrolysates contained high amounts of
monosaccharides (Glc, Xyl, and Man) which significantly obstruc-
ted identification of the used binder in painted samples (Fig. S3).
The filtration step prior the MAH procedure helped to reduce
the amount of interferences from paper matrix (even though traces
of Ara, Xyl and Glc were visible). In combination with direct analysis
of undiluted hydrolysates, gum Arabic profile was apparent in
samples of watercolor-painted paper (see Fig. 5).
3.4. Classification of plant gums and watercolors
Qualitative approach is not very reliable since the actual
composition of an art or historical sample might be affected by
117
different factors: presence of other saccharide materials, biological
attack and effect of aging [5,6]. For example, Xyl and Man can fail as
markers for fruit tree gums as they might originate from softwood.
Man can be derived from an egg binder. Rha and uronic acids
degrade in the presence of certain pigments. Glc and Fru are
commonly found in environment and may contaminate a sample.
In general, direct quantitative approach was not successful.
Absolute peak areas exhibited large variations mainly due to non-
homogenous nature of samples in case of plant gums (thus
different absolute amount of sugars in individual samples) or dif-
ferences in manufacturing of individual watercolors. Using relative
peak areas, RSD for plant gums decreased dramatically (Table S1). It
indicated constant ratios between individual monosaccharides.
RSD of relative peak areas for watercolors remained high even after
Glc exclusion. Since separation of Ara and Xyl was imperfect, their
summed peak areas and areas of well separated anomers were
used.
Since raw numbers were not sufficiently characteristic, mono-
saccharide peak area ratios (Rha to (AraþXyl), Fuc to (AraþXyl) and
Gal to (AraþXyl) [7]) were tested for material classification. All
three plant gums were distinguished from each other (Fig. S4).
Watercolor samples occupied position close to gum Arabic. How-
ever, paper, painted paper and filtered paper samples were not
distinguished and fell within the group of cherry gum samples. The
tested approach was appropriate for pure plant gums but was
inapplicable in the presence of other polysaccharide-based
materials.
PCA using monosaccharides relative peak area was more suc-
cessful in identification of pure materials. For the sake of compar-
ison, PCA including all variables (Fig. 6 a) and PCA without Glc
(Fig. 6 b) were performed. In both cases plant gum samples were
clearly distinguished. Alizarin crimson and Prussian blue over-
lapped with gum Arabic cluster. Hooker's green was out of this
position when all variables were used (influence of dextrin, section
3.3). After Glc exclusion Hooker's green merged with other water-
colors. Samples of hydrolyzed paper and watercolor-painted paper
were not distinguished regardless of Glc exclusion. However, when
the same samples were filtered before MAH and analyzed undi-
luted a clear classification was achieved and watercolors applied
onto paper merged with the cluster of gum Arabic.
118 V. Pauk et al. / Analytica Chimica Acta 989 (2017) 112e120

Fig. 4. Influence of MAH temperature on absolute peak area of individual monosaccharides from different samples. Watercolors bars combine data for Alizarin crimson, Hooker's
Green and Prussian Blue containing Arabic gum (Glc from dextrin was only found in Hooker's Green). Chromatographic conditions: C18SB, gradient: 0e30% methanol/water/formic
acid (91:5:4, v/v/v) in 5 min, 2.0 mL min1. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

3.5. Method characteristics and comparison
Our procedure offers the comparable performance for identifi-
cation of plant gums and has several important benefits that
overcome limitations or inconveniences of the reported methods.
Comparing to the previous SFC reports, our UHPSFC/MS method
covers more monosaccharides (e.g. Fuc, GlcA and GalA were never
mentioned in SFC works), achieves better separation of critical pairs
(Ara-Xyl and Glc-Gal were not separated in these works), resolves
other overlapping peaks by MS and allows for analysis of neutral
and acidic analytes in one run. Comparing to other techniques (see
Introduction), our procedure suggests dramatic reduction in labor,
simplifies and speeds up the sample preparation since it does not
require derivatization or sample cleanup (except filtration), allows
direct analysis of hydrolysates and has much shorter analysis time.
We have established second order polynomial calibration model
for neutral monosaccharides and linear model for uronic acids with
satisfactory characteristics. UHPSFC-MS method precision and ac-
curacy were better than 15% (20% at LOQ level) for all substances in
the range from 1  106 to 5  105 mol L1, which is satisfactory
for ESI-MS detection. RSD of anomer peak ratios were better than
3% for Gal, 6% for Ara and Xyl, and 14% for Fuc. Therefore, deter-
mination based on peak areas of well-separated anomers is
possible in case of overlapping saccharides. MAH procedure
 V. Pauk et al. / Analytica Chimica Acta 989 (2017) 112e120
Fig. 5. Chromatograms of hydrolysates: a) paper painted with Prussian blue water-
color, filtered before MAH, b) pure gum Arabic, 20 mg, c) paper, filtered before MAH.
Chromatographic conditions as in Fig. 4, hydrolysates were analyzed undiluted. (For
interpretation of the references to colour in this figure legend, the reader is referred to
the web version of this article.)
recovery was near-theoretical (101%, RSD < 8%), comparing to 72%
in the original report [8].
LODs of our method estimated based on S/N ratios
(0.01e0.12 ng mL1) were comparable with HPAEC-PAD
(0.01e0.03 ng mL1 [8]) and two-three orders of magnitude better
than those for GC/MS methods [5,13] (Table S2). Sample size for
identification of plant gum in a painting medium ranges from 0.1 to
15 mg [4,7e10,12,13]. Owing to high efficiency hydrolysis, low
Fig. 6. PCA plots based on relative peak areas of monosaccharides as variables, a) all variables, PC1 and PC2 account for 40.5% and 18.3% of the total variance, respectively; b) Glc
excluded, PC1 and PC2 account for 47.8% and 20.6% of the total variance. Chromatographic conditions as in Fig. 4. (For interpretation of the references to colour in this figure legend,
the reader is referred to the web version of this article.)
119
degradation of monosaccharides and possibility of direct analysis of
undiluted hydrolysates, sensitivity of our procedure was sufficient
for identification of 1 mg of gum Arabic (Fig. 7) which would
correspond sample size 0.1 mg for binder content 1%.
Our qualitative findings are in agreement with the GC-MS and
CE-UV literature data [3e6,10]. The absolute content of mono-
saccharides is rarely presented in the literature because calibration
dependences should be established for each substance in appro-
priate matrix, which is not always feasible. Results are often re-
ported as monosaccharide peak areas or their relative percentage
content. Numerical values of relative percentage content presented
here (Table S1) cannot be compared directly with literature data
because a different detection technique (ESI) was used. Further,
different response factors were observed for analytes. For instance,
relatively low peak areas of uronic acids in our results are explained
by their weaker response.
4. Conclusion
A new UHPSFC-MS method was developed for analysis of native
monosaccharides. Its performance was demonstrated during
analysis of plant gum binders in a painting medium as one of the
potential applications. Proper selection of mobile phase modifier
and additives was essential for satisfactory separation. Methanol
modifier with 5% water and 4% formic acid provided good peak
shapes for nine underivatized monosaccharides and elution of very
polar uronic acids in one chromatographic run. Among the tested
columns C18SB offered sufficient separation of studied substances
in 4.5 min. Anomer separation, often seen as a drawback, improved
identification of critical pairs Ara-Xyl and Glc-Gal not separated by
other SFC methods. Comparing to previous SFC reports, our method
also allowed direct injection of aqueous samples and covered all
plant gum constituents and two potential contaminants. High-
pressure MAH enabled fast release of monosaccharides from
plant gums and watercolor samples using 2 mol L1 TFA. 120  C
temperature represented the useful compromise between hydro-
lysis efficiency and yield of labile monosaccharides from all plant
gums and watercolor samples.
Two classification approaches and qualitative data were
compared for identification of a plant gum itself, in watercolors and
in watercolors on interfering paper support, which was barely re-
ported in analytical methods. Peak area ratios failed to distinguish
120 V. Pauk et al. / Analytica Chimica Acta 989 (2017) 112e120
Fig. 7. Chromatograms of undiluted hydrolysates: a) 1 mg sample of gum Arabic, b) blank, 2 mol L1 TFA.
cherry gum and paper hydrolysates. PCA with selected mono-
saccharide relative peak areas as variables successfully classified
pure materials, confirmed Arabic gum in watercolors and recog-
nized the binder in watercolors applied onto paper.
The developed method offers significant time and labor saving
and can serve as a simple and fast alternative to existing methods
for analysis of saccharide-based binders using e.g. HPAEC, CE and
GC. Owing to its conformity with requirements for analysis of real
samples, it might be useful but not limited to evaluation of painting
media. This work supports the extension of polarity range of SFC
applications and shows that SFC may compete with other chro-
matographic methods in the field of monosaccharides analysis.
Acknowledgement
The authors gratefully acknowledge the support by the project
LO1305 of the Ministry of Education, Youth and Sports of the Czech
Republic.
Appendix A. Supplementary data
Supplementary data related to this article can be found at http://
dx.doi.org/10.1016/j.aca.2017.07.036.
References
[1] A. Nussinovitch, Plant Gum Exudates of the World: Sources, Distribution,
Properties, and Applications, CRC Press, Boca Raton, FL, 2010, p. 401.
[2] B.A. Ormsby, J.H. Townsend, B.W. Singer, J.R. Dean, British watercolour cakes
from the eighteenth to the early twentieth century, Stud. Conserv. 50 (2005)
45e66.
[3] M.P. Colombini, A. Andreotti, I. Bonaduce, F. Modugno, E. Ribechini, Analytical
strategies for characterizing organic paint media using gas chromatography/
mass spectrometry, Acc. Chem. Res. 43 (2010) 715e727.
[4] M. Großl, S. Harrison, I. Kamla, E. Kenndler, Characterisation of natural poly-
saccharides (plant gums) used as binding media for artistic and historic works
by capillary zone electrophoresis, J. Chromatogr. A 1077 (2005) 80e89.
[5] H. Brecoulaki, A. Andreotti, I. Bonaduce, M.P. Colombini, A. Lluveras, Charac-
terization of organic media in the wall-paintings of the “Palace of Nestor” at
Pylos, Greece: evidence for a secco painting techniques in the Bronze Age,
J. Archaeol. Sci. 39 (2012) 2866e2876.
[6] A. Lluveras-Tenorio, J. Mazurek, A. Restivo, M.P. Colombini, I. Bonaduce, The
development of a new analytical model for the identification of saccharide
binders in paint samples, PLOS ONE (2012) e49383, 1e17.
[7] C. Riedo, D. Scalarone, O. Chiantore, Advances in identification of plant gums
in cultural heritage by thermally assisted hydrolysis and methylation, Anal.
Bioanal. Chem. 396 (2010) 1559e1569.
[8] M.P. Colombini, A. Ceccarini, A. Carmignani, Ion chromatography character-
ization of polysaccharides in ancient wall paintings, J. Chromatogr. A 968
(2002) 79e88.
[9] C. Riedo, D. Scalarone, O. Chiantore, Multivariate analysis of pyrolysis-GC/MS
data for identification of polysaccharide binding media, Anal. Mehods 5 (2013)
4060e4067.
[10] A. Lluveras-Tenorio, J. Mazurek, A. Restivo, M.P. Colombini, I. Bonaduce,
Analysis of plant gums and saccharide materials in paint samples: comparison
of GC-MS analytical procedures and databases, Chem. Cent. J. 6 (2012) 115.
[11] I. Bonaduce, H. Brecoulaki, M.P. Colombini, A. Lluveras, R. Vincenzo,
E. Ribechini, Gas chromatographicemass spectrometric characterisation of
plant gums in samples from painted works of art, J. Chromatogr. A 1175
(2007) 275e282.
[12] M. Mazanek, I. Kaml, E. Kenndler, Capillary electrophoresis: an alternative to
chromatography for analysis of natural organic binders, Stud. Conserv. 51
(2006) 139e151.
[13] A. Lluveras, I. Bonaduce, A. Andreotti, M.P. Colombini, GC/MS analytical pro-
cedure for the characterization of glycerolipids, natural waxes, terpenoid
resins, proteinaceous and polysaccharide materials in the same paint micro-
sample avoiding interferences from inorganic media, Anal. Chem. 82 (2010)
376e386.
[14] G.K. Webster (Ed.), Supercritical Fluid Chromatography. Advances and Ap-
plications in Pharmaceutical Analysis, Pan Stanford Publishing, Boca Raton,
2014, p. 394.
[15] T.A. Berger, Separation of polar solutes by packed column supercritical fluid
chromatography, J. Chromatogr. A 785 (1997) 3e33.
[16] T.A. Berger, Supercritical Fluid Chromatography, Primer, Agilent Technologies,
Inc., 2015.
[17] M.F. Mazzobre, M.V. Roman, A.F. Mourelle, H.R. Corti, Octanolewater partition
coefficient of glucose, sucrose, and trehalose, Carbohydr. Res. 340 (2005)
1207e1211.
[18] E. Lesellier, C. West, The many faces of packed column supercritical fluid
chromatography e a critical review, J. Chromatogr. A 1382 (2015) 2e46.
[19] A. Periat, A. Grand-Guillaume Perrenoud, D. Guillarme, Evaluation of various
chromatographic approaches for the retention of hydrophilic compounds and
MS compatibility, J. Sep. Sci. 36 (2013) 3141e3151.
[20] M. Ashraf-Khorassani, L.T. Taylor, E. Seest, Screening strategies for achiral
supercritical fluid chromatography employing hydrophilic interaction liquid
chromatography-like parameters, J. Chromatogr. A 1229 (2012) 237e248.
[21] L.T. Taylor, Packed column supercritical fluid chromatography of hydrophilic
analytes via water-rich modifiers, J. Chromatogr. A 1250 (2012) 196e204.
[22] M. Lafosse, B. Herbreteau, L. Morin-Allory, Supercritical fluid chromatography
of carbohydrates, J. Chromatogr. A 720 (1996) 61e73.
[23] A. Salvador, B. Herbreteau, M. Lafosse, M. Dreux, Subcritical fluid chroma-
tography of monosaccharides and polyols using silica and trimethylsilyl col-
umns, J. Chromatogr. A 785 (1997) 195e204.
sek, E. Búriov
[24] P. Brúder, F. Maca a, Electrospray ionization mass spectra of
pentoses, hexoses, and 2-deoxy-2-fluoro-D-glucose, Chem. Pap. - Chem.
Zvesti 62 (2008) 547e552.
[25] B.E. Poling, J.M. Prausnitz, J.P. O'Connell, The Properties of Gases and Liquids,
fifth ed., McGraw-Hill, New York, 2011, p. A.6.
[26] G. Karlsson, Development and application of methods for separation of car-
bohydrates by hydrophilic interaction liquid chromatography, in: P.G. Wang,
W. He (Eds.), Hydrophilic Interaction Liquid Chromatography (HILIC) and
Advanced Applications, CRC Press, Boca Raton, 2011, pp. 491e522.
[27] L. Bhattacharyya, J.S. Rohrer (Eds.), Applications of Ion Chromatography for
Pharmaceutical and Biological Products, first ed., John Wiley & Sons, 2012,
p. 455.
[28] D.R. Lide (Ed.), CRC Handbook of Chemistry and Physics Vol. 90, CRC Press,
Boca Raton, 2010.
kova , A.G. Perrenoud, I. Francois, C. West, E. Lesellier, D. Guillarme,
[29] L. Nova
Modern analytical supercritical fluid chromatography using columns packed
with sub-2 mm particles: a tutorial, Anal. Chim. Acta 824 (2014) 18e35.